On 14/06/2017 14:14, Roman Kagan wrote: >>> vcpu_id risks being confused KVM's vcpu_id (which is the CPU APIC >>> ID in x86). If you are already touching this code, this could be >>> renamed to vp_index to avoid confusion. >> Actually the VP_INDEX is _also_ the vcpu_id. Should we just document >> that KVM makes the VP_INDEX equal to the CPU APIC id, and adjust patch 5 >> accordingly? > I think here it makes sense to follow Eduardo's suggestion because this > code doesn't need to know how VP_INDEX is defined.
Yes, I agree. Sorry if that wasn't clear---Eduardo's comment just made me notice this possible solution to the issue of patch 5. Paolo > The real matter is discussed in patch 5, yes.
